Thesis idea: Speaker distortion detection
I had an idea for my thesis (just tonight so its still just an early idea), to design an amplifier that detects when speakers are distorting and limit the volume. Thus you would notice a max volume difference between good/crap headphones or speakers.
Anyone know of a system that already does this?
Anyone have a example waveform of a speaker in normal operation vs distortion?
I would just monitor the voltage waveform or current and compare it to a sample output.
I am not quite sure what is the main goal of such a thesis.
Do you want to
1. protect the speakers?
2. avoid distortion for the listeners?
3. actively compensate distortions?
1. For peaks a simple limiter will do. For overpower a thermal model of the connected speakers is required to prevent any case of destruction.
2. This can be done by limter or look ahead limiter (see below).
3. Is very complicated or even unreachable....
From DSP based controllers I once implemented a technique of a look ahead peak limiter. The audio signal is delayed by some samples, a limiter function detects voltages peaks beyond the maximum output voltage level and then reduces the amplitude before the actual peak reaches the output of the speakers. That way you have no hard clipping but a smooth attack. In analogue systems the peak amplitude is eigther strongly compressed to avoid peak overvoltage or with a smooth characteristic the first peak is always passed through and further peaks are limited.
I like the idea!
So small, bad cheapy loudspeakers would only play at low sound level, but large good quality loudspeakers would be alowed to player louder, because they distors less easely.
Now, I don't know what the non-linearity of the conus has for effect on the impedance, but I expect it to have some effect, and thus, it must be detectable..
But since the impedance of a loudspeaker can be complicated allready, maybe it's not practible to detect distortion from loudspeakers..
This is done a a number of subwoofers with servo control. Velodyne, Paradigm Reference, and Rythmik Audio
The Velodyne and Paradigm use an accelerometer hooked to the cone and the Rthmik uses an additonal sensing coil in the motor.
Philips introduced this in 1976
It was discussed somewhere at speakerplans.com - IIRC Void amps detect speaker's back EMF, which changes when x-max is exceeded, and limits output accordingly.
Thanks for the input!
I guess the main idea is to protect the speakers and all the hardware, and well just stop the bad practice of people cranking up the volume as far as it will go.
My line of thinking was that when a speaker reaches the end of its travel either the excess energy that would want to drive the speaker further would be visible on either the voltage or current waveform as a distortion, and then I would detect it and turn down the volume. This would be a built in feature of the amplifier so connecting up different speakers would have different limiting.
So as a basic set it up im thinking about using an op amp with feedback from a set of dummy transistors and using the same opamp output to drive the real transistors. That way the voltage on the speaker outputs is not used as feedback, and the output of the dummy transistors can double as the 'control' signal for comparing the speaker voltage to. Keeping both sets of transistors on the same heatsink should keep and temperature variations to a minimum. Hope this makes sense? I'll do a 2min paint diagram. Sure its a basic design but for proof of concept.
Further comments appreciated
If the amp is remote from the speakers, how would you compensate for different cross-overs and cable types/lengths?
as mentioned a way that has been utilized to do exactly what you are saying is the back emf.
Mind you most people damage their speakers not from too much power but from constantly clipping an amplifier of too little. While a more common issue in car audio than home audio, its long been known and said that most bass nut car audio boys blowing their subs with thei 150 dollar 2000 watt amplifier did so because they were clipping a 150 watt amplifier. The clipped waveform increases the heat in the voicecoil while not moving the cone enough for proper cooling, and thus, a blown speaker from too little power. Tweeters will blow most easily from being sent a constant clipped signal, and while tweeters don't need as much power for a given output typically (They often are more efficient than the midbass they are used with), its still a very real and common problem. A lot of tweeters can sustain a short duration blip of in excess of 100's and even 1000's of watts without damage, if its clean power (note that morel rates many of their tweeters for short term power of 1000 watts). Recent testing I did with my focal tweeter modifications indicated to me that the tweeter heated up at a rate beyond anything I could replicate with clean power, when it was fed a clipped signal. This is what fried the voicecoil in my tweeters the fastest. Feeding the tweeter 200 clean watts for 10 seconds with a 1st order crossover starting at 5khz indicated only a 5 degree F change over ambient at the pole piece, no increase in distortion in the before and after tests. However, measuring the same tweeter with an amplifier that clips at over 15% distortion (IMD) after 18 watts or so, and feeding a square wave signal into that amplifier as well caused the tweeter to blow in that 10 second period. I don't remember the temp rise, but it was drastically higher, like pole piece air temp changed 60 degrees or something like that.
When an amplifier is clipping the music signal it will generate additional harmonics.
E.g. you have a pure sinus input to your amplifier of 100 Hz and increase the amplitude that way, that the output almost looks like a square wave. This output will be a 100Hz square signal containing a lot odd harmonics like 3,5,7,9,11,13,15,17,19. A tweeter for example used at 1kHz will play all the higher harmonics. 11th harmonics of 100Hz -> 1,1kHz. In total the tweeter has to withstand the total energy of all harmonics higher that lets say 7th/9th order.
This has two effects:
1. The tweeter will die because of overheating the coil as energy ment for the bass speaker is shifted due to clipping in higer frequency regions.
2. Especially high energy arround the resonance frequency of the tweeter (mostly below the frequency you can start using a tweeter) the peak exitation (movement) of the diaphragma will be that high, that I can break easily. For this maybe only 1/10 th of the rated power is needed.
|All times are GMT. The time now is 08:39 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.
Copyright ©1999-2016 diyAudio